YouTube Comedy Slam Preference Data
Donated on 4/9/2012
This dataset provides user vote data on which video from a pair of videos is funnier collected on YouTube Comedy Slam. The task is to automatically predict this preference based on video metadata.
Dataset Characteristics
Text
Subject Area
Computer Science
Associated Tasks
Classification
Feature Type
-
# Instances
1138562
# Features
3
Dataset Information
Additional Information
YouTube Comedy Slam (http://www.youtube.com/slam/comedy) is a video discovery experiment running on YouTube's version of labs (called TestTube) for a few months in 2011 and 2012. In this experiment, a pair of videos were shown to the user and the user was asked to vote for the video that they found funnier. Left/right positions of the videos were randomly selected before being presented to the user to eliminate position bias. Videos were selected from a large pool of weekly updated sets of videos. One of the outcomes of this experiment is a training dataset for automatically predicting which video would be deemed funnier by users using a variety of features. For example, uploader supplied metadata and/or user comments on watch pages of these videos could be used as features. See http://googleresearch.blogspot.com/2012/02/quantifying-comedy-on-youtube-why.html for more detail. The attached dataset includes roughly 1.7 million preference votes. The votes were recorded chronologically. The first 80% are provided here as the training dataset and the remaining 20% as the testing dataset. Each line in this dataset corresponds to one vote over a pair of YouTube videos. Each video is represented by its YouTube video ID. For example, the watch page URL for video ID 'txqiwrbYGrs' is http://www.youtube.com/watch?v=txqiwrbYGrs. User preference over a pair of videos is presented in the form of string “left†if the left video was deemed funnier, and “right†otherwise. This user vote should be used as ground truth for both training and testing. Evaluation should be based on average accuracy in predicting this preference over the testing partition (provided). Any other information about the vote (e.g. ID of the user) is not provided.
Has Missing Values?
No
Variable Information
Each row in this text file represents one anonymous user vote. Each line contains three comma-separated fields. The first two fields are YouTube video IDs. The third field is either 'left' or 'right'. Left indicates the first video from the pair was voted to be funnier than the second. Right indicates the opposite preference.
Dataset Files
| File | Size |
|---|---|
| comedy_comparisons.train | 25.7 MB |
| comedy_comparisons.test | 6.4 MB |
Reviews
There are no reviews for this dataset yet.
pip install ucimlrepo
from ucimlrepo import fetch_ucirepo # fetch dataset youtube_comedy_slam_preference_data = fetch_ucirepo(id=223) # data (as pandas dataframes) X = youtube_comedy_slam_preference_data.data.features y = youtube_comedy_slam_preference_data.data.targets # metadata print(youtube_comedy_slam_preference_data.metadata) # variable information print(youtube_comedy_slam_preference_data.variables)
Shetty, S. (2012). YouTube Comedy Slam Preference Data [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C5102Z.
Creators
Sanketh Shetty
DOI
License
This dataset is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.
This allows for the sharing and adaptation of the datasets for any purpose, provided that the appropriate credit is given.